*Not like this robot-vision stuff is hard work by engineers, or anything. Efficiently solving multi-label MRFs (Readme) (C/C++ code) Segmentation, object category labelling, stereo Multi-label ...